Local Drug Delivery in the Treatment of Glioblastoma

The prognosis following diagnosis of glioblastoma remains poor. Historically, there have been many notable attempts to use local drug delivery to treat glioblastoma, including convection-enhanced delivery (CED), direct tumor injection, and the use of to deliver chemotherapeutics. The use of polymer wafers resulted in the only US Food and Drug Administration (FDA)–approved intracranial drug implant for treatment of recurrent and de novo glioblastoma, Gliadel. Over a period of several decades, the delivery of the chemotherapeutic 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) from biodegradable polyanhydride polymeric wafers was developed and tested for efficacy and safety, culminating in the wafer’s approval for recurrent glioblastoma in 1996 and de novo glioblastoma in 2003. To date, this polymeric delivery vehicle is the most extensively tested and also one of the few drug-eluting systems created for intracranial drug delivery. This chapter reviews the scientific challenges and evolution of local compounds for glioblastoma, including BCNU wafers, CED, direct injection, and hydrogel-based delivery.

Background to intracranial drug delivery

The impetus for local drug delivery is fundamentally based on the ability to bypass the blood-brain barrier (BBB). The BBB is composed of endothelial cells in brain capillaries forming nearly impenetrable tight junctions (zonula occludens), with limited to almost no detectable pathways for transendothelial transport of molecules of greater than 200 Da. Beyond a physical barrier, the BBB also forms a metabolic gate, inactivating passive compounds via intracellular and extracellular enzymes. Elsewhere in this publication, techniques and modalities specialized for bypassing the BBB for more efficacious systemic administration of drugs are described. More specifically, this chapter addresses the tools, materials, and techniques used in the direct bypass of the BBB by local, intracranial drug delivery.

The implantation of drug-eluting biomaterials within the brain is limited by several factors. The first and most important is that any implantable material must maintain a highly favorable safety profile. All implantable devices within the brain parenchyma are subject to interaction between the device and brain tissue and must be (1) completely biocompatible without eliciting a biological reaction, (2) unable to migrate and cause mechanical damage to the ventricular system or anatomic structures, and (3) limited in its ability to cause severe side effects from leakage of the dissolving or migrating compound.

A major barrier to the implementation of local and topical drugs for the treatment of glioblastoma is the cellular infiltrative nature of the disease, which extends beyond where a resection may take place surgically, or where a topical agent can be applied. Local drug delivery can provide controlled, sustained release of the compound, or can enhance the permeability of the BBB in some cases to promote uptake of the therapeutic, but does not change the underlying pathophysiologic nature of diffuse cellular infiltration.

Convection-enhanced delivery

On the basis that intratumoral drug administration is limited by poor diffusion of drugs through brain interstitium, CED promotes improved diffusion and distribution of small and large molecules via establishment and maintenance of a pressure gradient during interstitial infusion. This improvement is accomplished by administering a controlled pressure differential via interstitial infusate. Practically, this is accomplished by stereotactically placing a cannula or microcatheter near to or around the tumor cavity through a burr hole in the skull. This catheter is connected to a mechanized pumping device that creates a pressure gradient through a constant infusion rate, concentration, and duration of administration. The system has several advantages. The first is that it allows bypass of the BBB, and allows larger molecules to be administered within whatever volume of parenchyma is being perfused by CED. This ability may manifest on the order of kilodaltons or greater, compared with the 200 Da that are able to permeate the BBB without convection. The distribution of infusates can also be targeted, and they are generally thought to be more homogeneously distributed than with other delivery approaches. Further, the effective diffusion distance from initial microcatheter placement can be as much as 3 cm, compared with the several millimeters that can be achieved via continuous local infusion without CED, or with topical drug placement without convection. The downsides to CED are that smaller, hydrophilic molecules are more likely to leak out into the interstitial tissue from the central nervous system vasculature and into the systemic circulation ( Fig. 16.1 ).

The clinical use of CED has been wide-ranging. In addition to conventional chemotherapeutics, the delivery of a variety of toxins, antibodies, and vectors has been attempted. The best clinical evidence regarding the use of CED is the phase III PRECISE trial. CED of cintredekin besudotox (a recombinant chimeric cytotoxin of human interleukin-13 fused to a mutated form of Pseudomonas exotoxin that kills tumor cells expressing the interleukin 13 receptor overexpressed by malignant glioma cell lines) was compared with Gliadel wafers in patients with their first recurrence of glioblastoma. A total of 296 patients were enrolled at 2 centers and were randomized to either CED of the toxin or Gliadel wafers. The overall survival from time of randomization was the primary end point. Median overall survival was 9.1 months for CED and 8.8 months for Gliadel wafers ( P = .476). In addition to being nonsuperior, the pulmonary embolism rate was higher in the CED group (8% vs 1%). As described, the study may have been flawed by an inadequate percentage of catheter placements being performed per protocol specifications and stringent requirements for showing survival benefit. It should also be noted that there was an improvement in progression-free survival (17.7 weeks for CED vs 11.4 weeks for Gliadel wafer) but this was considered post hoc and not a primary or secondary end point by the study investigators.

In past few years, additional agents, such as liposomes and nanoparticles, have been the focus of preclinical investigation with CED. Advances in catheter placement and technique, as well as visualization of convection with real-time imaging, are improving preclinical promise, and both phase I and phase II trials are underway and may progress to phase 3 over the next few years.

Drug-eluting wafers

The Gliadel wafer (carmustine wafer [CW]) is by far the most well established and well studied of the local drug-eluting systems for intracranial delivery, and the only FDA-approved compound therapy for use in glioblastoma apart from temozolomide and bevacizumab. The development of this product originated from work by Robert Langer and Judah Folkman in the 1970s on the use of polymers for sustained release of proteins and macromolecules. The biocompatibility of these delivery systems was then tested and evaluated throughout the 1980s for both safety and efficacy in preclinical models by Henry Brem and his team at Johns Hopkins Hospital. As described by Brem and Gabikian, the CW (BCNU) was integrated into a controlled delivery polymer (poly[bis(p-carboxyphenoxy) propane with sebacic acid) for controlled release. As a chemotherapeutic, BCNU functions as an alkylating agent to inhibit the synthesis of DNA, RNA, and other proteins. Its efficacy during systemic distribution is hampered by severe toxicities that include bone marrow suppression and cytotoxic effects on end-organ systems. The integration of BCNU into a controlled-delivery polymer both circumvented the BBB and allowed sustained release of high concentrations of BCNU directly into the tumor resection cavity, with sustained release that was found 2 weeks or longer after implantation.

After successful animal studies, a phase I to II trial of 21 patients with recurrent glioblastoma was performed with CW treatment in 1981, which showed successful release of BCNU with no adverse effects to the study population. This study was followed in 1995 by a randomized, placebo-controlled prospective study of CW in recurrent glioblastoma. A total of 222 patients in 27 medical centers were randomized to either receive functioning CW or placebo wafers. The median survival of the 110 patients who received CW was 31 weeks versus 23 weeks for the 112 patients who received a placebo. During the study there were no clinically noted adverse events attributable to the CW. Long-term follow-up data of this patient population were published in 2003, which included additional patients with de novo glioblastoma and reported a sustained survival benefit of 13.9 months for patients receiving the CW and 11.6 months in the placebo group. However, at that time adverse events attributable to CW were significantly greater when accounting for cerebrospinal fluid leaks and intracranial hypertension. As a result of these studies, and after FDA review, the CWs now known as Gliadel gained FDA approval for recurrent glioblastoma in 1996, and de novo glioblastoma in 2003 ( Fig. 16.2 ).